next up previous contents
Next: Q3 - Local Dependence Up: Benchmarking HPJava, Part II: Previous: Laplace Equation Using Red-Black   Contents


3-Dimensional Diffusion Equation

Figure 7.5: 3D Diffusion Equation on shared memory machine
Figure 7.5 shows the performance of the 3D diffusion on the shared memory machine. Again, we need to see the Java performance over the C performance on the shared memory machine. It is 104.33% over C. The table 7.9 shows the speedup of the HPJava naive translation over sequential Java and C programs. Moreover, it shows the speedup of HPJOPT2 over the naive translation.


Table 7.9: Speedup of the naive translation over sequential Java and C programs for the 3D Diffusionequation on the shared memory machine.
Number of Processors 1 2 3 4 5 6 7 8
Naive translation
over Java 0.54 1.00 1.35 1.88 2.09 2.73 2.99 3.42
HPJOPT2 over Java 1.05 2.02 2.66 3.59 4.02 4.32 5.18 5.98
HPJOPT2 over
Naive translation 1.97 2.02 1.97 1.91 1.93 1.58 1.73 1.75

The speedup of the naive translation with 8 processors over sequential Java is up to 342%. The speedup of HPJOPT2 with 8 processors over sequential Java is up to 598%. The speedup of HPJOPT2 over the naive translation is up to 175%. The table 7.10 shows the speedup of the naive translation and HPJOPT2 for each number of processors over the performance with one processor.


Table 7.10: Speedup of the naive translation and HPJOPT2 for each number of processors over the performance with one processor for the 3D Diffusion equation on the shared memory machine.
Number of Processors 2 3 4 5 6 7 8
Naive translation 1.87 2.52 3.52 3.90 5.10 5.58 6.39
HPJOPT2 1.92 2.52 3.41 3.82 4.10 4.92 5.67

Figure 7.6: 3D Diffusion Equation on distributed memory machine
\includegraphics[width=5in]{Figures/diff3d_128_SP3}
The naive translation gets up to 639% speedup using 8 processors on the shared memory machine. Moreover, HPJOPT2 gets up to 567%. Figure 7.6 shows performance for the 3D diffusion equation on the distributed memory machine. Table 7.11 also shows the speedup of the naive translation for each number of processors over the performance with one processor for the 3D diffusion equation on the machine.


Table 7.11: Speedup of the naive translation for each number of processors over the performance with one processor for the 3D diffusion equation on the distributed memory machine.
Number of Processors 4 9 16 32
Naive translation 3.31 5.76 9.98 13.88


next up previous contents
Next: Q3 - Local Dependence Up: Benchmarking HPJava, Part II: Previous: Laplace Equation Using Red-Black   Contents
Bryan Carpenter 2004-06-09